Memory Management. Reading: Silberschatz chapter 9 Reading: Stallings. chapter 7 EEL 602

Transcription

1 Memory Management Reading: Silberschatz chapter 9 Reading: Stallings chapter 7 1

2 Outline Background Issues in Memory Management Logical Vs Physical address, MMU Dynamic Loading Memory Partitioning Placement Algorithms Dynamic Partitioning Buddy System Paging Memory Segmentation Example Intel Pentium 2

3 Background Main memory fast, relatively high cost, volatile Secondary memory large capacity, slower, cheaper than main memory and is usually non volatile The CPU fetches instructions/data of a program from memory; therefore, the program/data must reside in the main (RAM and ROM) memory Multiprogramming systems main memory must be subdivided to accommodate several processes This subdivision is carried out dynamically by OS and known as memory management 3

4 Issues in Memory Management Relocation: Swapping of active process in and out of main memory to maximize CPU utilization Process may not be placed back in same main memory region! Ability to relocate the process to different area of memory Protection: Protection against unwanted interference by another process Must be ensured by processor (hardware) rather than OS Sharing: Flexibility to allow several process to access the same portions of the main memory Efficiency: Memory must be fairly allocated for high processor utilization, Systematic flow of information between main and secondary memory 4

5 Binding of Instructions and Data to Memory Compiler Generates Object Code Linker Combines the Object code into a single self sufficient executable code Loading Copies executable code into memory Execution dynamic memory allocation 5

6 Binding of Instructions and Data to Memory Address binding of instructions and data to memory addresses can happen at three different stages Compile time: If memory location known a priori, absolute code can be generated; must recompile code if starting location changes Load time: Must generate relocatable code if memory location is not known at compile time Execution time: Binding delayed until run time if the process can be moved during its execution from one memory segment to another most general purpose OS 6

7 Logical Vs Physical Address Space Each logical address is bound to physical address space; Logical address generated by the CPU; also referred to as virtual address Physical address address seen by the memory unit Logical and physical addresses ; Identical in compile-time and load-time addressbinding schemes Differ in execution-time address-binding scheme Logical address Virtual address 7

8 Memory-Management Unit (MMU) The runtime mapping from virtual physical address Relocation register is added to every address generated by user process The user program logical addresses, it never sees the real physical addresses 8

9 Dynamic Loading Routine is not loaded until it is called Better memory-space utilization unused routine is never loaded Useful to handle infrequently occurring cases, e.g. error handling routines No special support from the OS required implemented through user program design 9

10 Memory Partitioning Two schemes used in several variations of now-obsolete OS Fixed Partitioning: OS occupies fixed portion of main memory, rest available for multiple processes. Two alternatives; Equal size fixed partitions any process partition size can be loaded Unequal size partitions several unequal size partitions, process of matching sizes Problems with equal size fixed partitions: If program is bigger than a partition size, use of overlays Main memory utilization is extremely inefficient; Internal Fragmentation waste of space internal to partition due to the fact that block of data loaded is smaller than partition 10

11 Unequal-Size Partitions Assign each processes the smallest partition to which it will fit Advantages: Process are always assigned in such a way as to minimize wasted memory within a partition internal fragmentation Relatively simple and require minimal OS software and overhead Disadvantages: Limitations on the active number of processes, number of partitions specified at system generation time Small jobs cannot utilize partition space efficiently; In most cases it is an inefficient technique 11

12 Placement Algorithm with Partitions Equal-size partitions Because all partitions are of equal size, it does not matter which partition is used Unequal-size partitions Can assign each process to the smallest partition within which it will fit Queue for each partition size Processes are assigned in such a way as to minimize wasted memory within a partition 12

13 Placement Algorithm with Partitions 13

14 Dynamic Partitioning Developed to address the drawbacks of fixed partitioning Partitions of variable length and number; Process in bought into main memory, it is allocated exactly as much memory as it requires Leaves Holes First at the end eventually lot of small holes Memory becomes more fragmented with time, memory utilization External Fragmentation Memory that is external to all partitions becomes increasingly fragmented Compaction Used to overcome external fragmentation OS shifts processes so that free memory is together in one block Compaction requires use of dynamic relocation capability Time consuming procedure and wasteful of processor time 14

15 Dynamic Partitioning 15

16 Placement Algorithms Compaction is time consuming OS must be clever in plugging holes while assigning processes to memory Three placement algorithms Selecting among free blocks of main memory Best-Fit: Closest in size to the request First-Fit: Scans the main memory from the beginning and first available block that is large enough Next-Fit: Scans the memory from the location of last placement and chooses next available block that is large enough 16

17 Placement Algorithms - Example Allocation of 16 MB block using three placement algorithms 17

18 Placement Algorithms Which of the above approaches is the best? Process Size/Sequence, General Comments First-Fit Simplest, usually the best and fastest Next-Fit Slightly worst results with next fit Compaction may be more frequently required Best-Fit Usually the worst performer; main memory is quickly littered by blocks too small to satisfy memory allocation requests Compaction - more frequently than other algorithms 18

19 Buddy System Drawbacks Fixed partitioning: Limits number of active process, inefficient if poor match between partition and process sizes Dynamic Partitioning: Complex to maintain, includes the overhead of compaction Compromise may be the Buddy System - Entire space available is treated as a single block of 2 U If a request of size s such that 2 U-1 < s 2 U, entire block is allocated Otherwise block is split into two equal buddies Process continues until smallest block greater than or equal to s is generated 19

20 Buddy System - Example Initial block size 1 MB; First request A is for 100 KB 20

21 Buddy System - Example Binary tree representation immediately after Release B request. 21

22 Relocation A process may occupy different partitions which means different absolute memory locations during execution (from swapping) Compaction will also cause a program to occupy a different partition which means different absolute memory locations 22

23 Paging Partitioning main memory small equal fixed-size chunks Each process is divided into the same size chunks pages Chunks of memory frames or page frames Advantages No external fragmentation Internal fragmentation only a fraction of last page of a process OS maintains a page table for each process Contains frame location for each page in the process Memory address a page number, a offset within the page Processor hardware logical-to-physical address translation 23

24 Paging - Example Assignment of process pages to free frames 24

25 Paging - Example Assignment of process pages to free frames. 25

26 Paging - Example Data structures for page tables at time epoch (f) 26

27 Paging - Example Convenience in Paging scheme Frame size power of 2 Relative address (wrt origin of program) and the logical address (page # and offset) are same Example - 16 bit address, page size 1K or 1024 bytes Maximum 64 (2 6 ) pages of 1K bytes each Advantages Logical addressing transparent to programmer, assembler, linker Relatively easy to implement a function to perform dynamic address translation at run time 27

28 Paging - Example 28

29 Paging - Example Logical-to-physical address translation in Paging 29

30 Paging - Example Logical-to-physical address translation in Paging 30

31 Implementation of Page Table Different methods of storing page tables, OS dependent Pointer to page table PCB Hardware implementation of page tables Page table Set of dedicated high speed registers, Simplest Suitable for small page table sizes, Usually very large requirements Page table is kept in main memory Page-table base register (PTBR) points to the page table Two memory access, page table and other for data/instruction Memory access slowed by a factor of two Solution to the two memory access problem Usage of a special fast-lookup hardware cache called associative memory or translation look-aside buffers (TLBs) TLB contains Page # Frame #, Small # of TLB entries ( ) 31

32 Paging Hardware With TLB 32

33 Shared Pages Shared code One copy of read-only (reentrant) code shared among processes, e.g. text editors, compilers Shared code must appear in same location in the logical address space of all processes Private code and data Each process keeps a separate copy of the code and data The pages for the private code and data can appear anywhere in the logical address space 33

34 Shared Pages Example Sharing of code in paging environment 34

35 Segmentation Memory-management scheme that supports user view of memory Program Collection of segments (name and length) Complier automatically constructs segments reflecting input program Example A C complier might create separate segments for the following main program, procedure, function, object, local variables, global variables, common block, stack, symbol table, arrays 35

36 Segmentation The program/process and its associated data is divided into a number of segments All segments of all programs do not have to be of the same length There is a maximum segment length Addressing consist of two parts - a segment number and an offset Since segments are not equal, segmentation is similar to dynamic partitioning 36

37 Address Translation Architecture 37

38 User s View of a Program 38

39 Logical View of Segmentation user space physical memory space 39

40 Example of Segmentation 40

41 Sharing of Segments 41

42 Segmentation Compared to dynamic partition, segmentation program may occupy more than one partition and these partitions need not be contiguous Segmentation eliminates the need for internal fragmentation but like dynamic partitioning it suffers from external fragmentation Process is broken in small pieces, the external fragmentation is less with segmentation than dynamic partition Paging is invisible to the programmer, segmentation is usually visible 42

43 Segmentation EXAMPLE: Logical Addresses. 43

44 Segmentation EXAMPLE: Logical-to-physical address translation in Segmentation 44

45 Hierarchical Page Tables Most systems support a large logical address space , page table itself becomes excessively large Break up the logical address space into multiple page tables A simple technique is a two-level page table 45

46 Two-Level Paging Example A logical address (32-bit machine with 4K page size) is divided into: a page number consisting of 20 bits a page offset consisting of 12 bits Since the page table is paged, page number is further divided into: a 10-bit page number a 10-bit page offset Thus, a logical address is as follows: page number page offset p i p 2 d where p i is an index into the outer page table, and p 2 is the displacement within the page of the outer page table 46

47 Two-Level Page-Table Scheme 47

48 Address-Translation Scheme Address-translation scheme for a two-level 32-bit paging architecture 48